Identification of novel prognostic biomarkers for thyroid cancer by integrated transcriptome analysis of metastasis-associated genes

Distant metastasis (DM) is the most important prognostic factor affecting overall survival (OS) of thyroid cancer. The current study aimed to discover prognostic biomarkers to predict thyroid cancer survival, particularly PTC, the most common subtype of thyroid cancer. Four RNA-seq datasets of experimental lung metastasis from 4 transgenic mouse models of PTC, follicular thyroid cancer (FTC), poorly-differentiated thyroid cancer (PDTC), and anaplastic thyroid cancer (ATC) were integrated to screen for candidate genes involved in DM. TCGA-THCA dataset were used to validate the candidate genes. A total of 105 up-regulated and 25 down-regulated differentially expressed genes (DEGs) were identified to be present in all 4 datasets. Regulation of cytokine production, inflammation, immune checkpoint regulation, MAPK/ERK cascade were major enriched pathways in metastatic tumor cells. We identified 7 genes whose overexpression was present in 63 of 498 PTC samples (13%) and was associated with poor OS (p<0.01). Clinically, the 7–gene expression signature was associated with older age at the diagnosis, late stage of tumor, tall-cell variant, and higher aneuploidy and hypoxia score. Mutation load was increased in samples with 7–gene expression signature: 26 samples had more than one driver mutations (47%, 26/55). Deep deletions in other chromosomal locus were frequently found in samples with BRAFV600E point mutations. In contrast, only 7% samples without the 7-gene expression signature had more than one driver mutations (24/243). Increased copy number variants (CNVs) were also observed in metastatic as compared to primary tumor cells such as large deletions and duplications. We conclude that the 7–gene expression signature is associated with poor prognosis and chromosomal instability. It may be a useful biomarker for risk stratification for DM and help decision-making in initial surgical recommendations.